A Multi - Agent Approach to Dynamic Traffic Assignment Based on Activity

The road choice behavior of travelers is closely related to their activity planning and location choice. This approach has led to the increasing development of multi-agent and activity-oriented modeling. This work attempts to model travelers’ dynamic departure time/route/destination choice behavior in a queuing network. In this respect, we propose an activity-based predictive dynamic traffic assignment model based on a multi-agent approach with two interacting levels: the travelers’ adaptive reaction level and the network propagation level. For the first level, traffic conditions change according to travelers’ departure time/route/destination choices, dynamic traffic information and network supply constraints. En-route dynamic traffic assignment is reflected by travelers’ strategies in response to the traffic state and the activity distribution information. For the second level, traffic congestion is modeled by point queue dynamics concept on a network. As a solution method of the predictive equilibrium, we propose an ACO (ant colony optimization) algorithm based on a time-dependent multi-type pheromone scheme in order to solve the proposed dynamic traffic assignment model. This algorithm focuses on how to provide dynamic on-route information and off-route information to guide travelers to make the best travel decision in a dynamic environment. The algorithm can be adapted to communication and information exchange schemes closer to actual traveler behaviour. A numerical example is given to illustrate the performance of the proposed method.